768k Skylake Problem/Bug

[kladner]

Originally Posted by chalsall

I'm not so sure. Easy to achieve that kind of load at idle spin with no seek.

Please also note this: "My last freeze (and the first AFTER I thought I was finally good) now was with high load on CPU and HDD".

Perhaps pegnose will speak to our questions.

Good point.

[kladner]

Originally Posted by chalsall

Just to put this out there...

Almost eleven years old, and yet still relevant....

Great article! Thanks!

[pegnose]

Ok, I don't get it. You mean resyncing two "Eco" drives with 5400 rmp each knocks out my gaming rig? Even if that consumes like 10+ times that much power on (actual) onload? As I said: I appreaciate your concern (I really do), but I don't think so. We are not talking about hiph-speed SCSI/SAS drives from outer space here.

EDIT: Regarding the power consumption of the HDDs, first I took it off the specs, then I measured via subtraction. Seems that they consume up to 6 W each on full load. Now let it be 12 W on spikes and - of both drives spiked synchronously - 24 W for both. That still is no big deal.

[Brunnis]

Originally Posted by pegnose

No. As I said: I started out with memory, and I am not yet done with it.

But, of course, part of the problem is - and this is more or less the same with all of us: I bought my mobo more than 6 months ago. What will my vendor say if I want to return it without being able to proove that it is broken and that it WAS broken from the beginning (after 6 mo that is necessary), PLUS that I want a different one in return? I should be happy if he deems my worthy of even the shortest response.

AND on the other hand: as I said, I don't believe that something is broken (or sort of broken with all ASUS Z170 boards). If my other componentes are fine, ASUS support has to deal with it. I will make them. Oh, I will.

EDIT: You HEAR me, ASUS?!?

Don't know if this is of any help, but below are the specs of my Skylake system which appears fully stable so far. I've had it for a couple of months now.

EVGA SuperNova G2 750W
ASUS Z170-A (currently BIOS 1602, but 1203, 1302 and 1402 gave me no issues)
2x8GB Corsair Vengeance LPX Black DDR4 2400MHz @ 2666MHz 16-16-16-39-1T @ 1.254V
Core i7-6700K (stock)
Sapphire Radeon R9 390 Nitro
ASUS Xonar DX (PCI-E soundcard)
Samsung 850 EVO 1TB
An old Samsung DVD-RW drive
Windows 10 Home x64 (Fully updated and clean install. No registry tweaks and no advanced power management profile modifications.)

The BIOS is mostly at default settings. I have changed the following:

- ASUS Multicore Enhancement: disabled (enabling this feature basically runs the CPU out of spec, although it "should" be harmless)
- FCLK: 1GHz (see http://www.anandtech.com/show/9607/s...-optimizations)
- Voltages manually set to the following, since I don't trust ASUS with "Auto" values:

- VCCIO: 0.95000V
- System Agent: 1.05000V
- PCH Core Voltage: 1.00000V
- CPU Standby Voltage: 1.000V

- PCI Express Native Power Management: enabled (I think this was actually enabled by default in earlier BIOS versions and I've kept it enabled since)
- HD Audio Controller: disabled
- Intel Thunderbolt: disabled

[pegnose]

Originally Posted by Brunnis

Don't know if this is of any help, but below are the specs of my Skylake system which appears fully stable so far. I've had it for a couple of months now.

EVGA SuperNova G2 750W
ASUS Z170-A (currently BIOS 1602, but 1203, 1302 and 1402 gave me no issues)
2x8GB Corsair Vengeance LPX Black DDR4 2400MHz @ 2666MHz 16-16-16-39-1T @ 1.254V
Core i7-6700K (stock)
Sapphire Radeon R9 390 Nitro
ASUS Xonar DX (PCI-E soundcard)
Samsung 850 EVO 1TB
An old Samsung DVD-RW drive
Windows 10 Home x64 (Fully updated and clean install. No registry tweaks and no advanced power management profile modifications.)

The BIOS is mostly at default settings. I have changed the following:

- ASUS Multicore Enhancement: disabled (enabling this feature basically runs the CPU out of spec, although it "should" be harmless)
- FCLK: 1GHz (see http://www.anandtech.com/show/9607/s...-optimizations)
- Voltages manually set to the following, since I don't trust ASUS with "Auto" values:

- VCCIO: 0.95000V
- System Agent: 1.05000V
- PCH Core Voltage: 1.00000V
- CPU Standby Voltage: 1.000V

- PCI Express Native Power Management: enabled (I think this was actually enabled by default in earlier BIOS versions and I've kept it enabled since)
- HD Audio Controller: disabled
- Intel Thunderbolt: disabled

Thanks for listing this so detailedly, Brunnis! Only, I am afraid the only thing our rigs have in commen ins the CPU. :-/

I have a few questions, though:

- Settings the voltages means the don't get reduced via c-states, right? You might be on to something, here.
- Did you configure your RAM manually, or did you load XMP, or a mixture of both?

[Brunnis]

Originally Posted by pegnose

Thanks for listing this so detailedly, Brunnis! Only, I am afraid the only thing our rigs have in commen ins the CPU. :-/

I have a few questions, though:

- Settings the voltages means the don't get reduced via c-states, right? You might be on to something, here.
- Did you configure your RAM manually, or did you load XMP, or a mixture of both?

No problem! Good questions. Let's see:

- I believe (but I'm actually not sure) that only the CPU core voltage is affected by the C-states. With my current settings (CPU voltage at default), the CPU voltage jumps around as expected depending on load. I have not paid attention to how the other voltages behave with BIOS stock settings, but maybe you have? I guess it should be possible to check with HWMonitor or similar. If they do jump around, I guess this could have an effect. I only used auto values for a short time after building the system.

- I have avoided loading XMP settings. I've simply set the memory to 2666MHz, kept every timing at default (auto), except CAS, RAS to CAS, RAS ACT Time and Command Rate, which I've set to 16, 16, 39 and 1. And then of course memory voltage to 1.25V. Actually not sure if the extra memory voltage is needed with my current settings, but it's not hurting anything. Since the memory is in fact overclocked, I thought I might as well leave it with 0.05V extra voltage to provide some additional margin.

[pegnose]

Originally Posted by Brunnis

No problem! Good questions. Let's see:

- I believe (but I'm actually not sure) that only the CPU core voltage is affected by the C-states. With my current settings (CPU voltage at default), the CPU voltage jumps around as expected depending on load.

Ok, but the lower limit is 1.0 V (your CPU standby voltage). This is interesting, because IIRC CPU core voltage drops as low as 0.8xy V with stock settings. This is what one guy reported at Tomshardware who could improve his situation with RMAing his 6700k. Maybe not all CPUs that passed the QM can actually stand this - and hence the lot of people with idle issues!

Originally Posted by Brunnis

I have not paid attention to how the other voltages behave with BIOS stock settings, but maybe you have? I guess it should be possible to check with HWMonitor or similar. If they do jump around, I guess this could have an effect. I only used auto values for a short time after building the system.

Not yet, but I'll have a close look!

Originally Posted by Brunnis

- I have avoided loading XMP settings. I've simply set the memory to 2666MHz, kept every timing at default (auto), except CAS, RAS to CAS, RAS ACT Time and Command Rate, which I've set to 16, 16, 39 and 1. And then of course memory voltage to 1.25V. Actually not sure if the extra memory voltage is needed with my current settings, but it's not hurting anything. Since the memory is in fact overclocked, I thought I might as well leave it with 0.05V extra voltage to provide some additional margin.

This sounds very reasonable. And many people over the web have reported that upping the DRAM voltage a bit helped their stability.

Cool, thanks for your info!!

[Brunnis]

Originally Posted by pegnose

Ok, but the lower limit is 1.0 V (your CPU standby voltage). This is interesting, because IIRC CPU core voltage drops as low as 0.8xy V with stock settings. This is what one guy reported at Tomshardware who could improve his situation with RMAing his 6700k. Maybe not all CPUs that passed the QM can actually stand this - and hence the lot of people with idle issues!

I'm not sure exactly when the CPU standby voltage is applied, but it's not the voltage used when idling in Windows. Even though I've set the standby voltage to 1.0V, the CPU is idling at 0.8V and 800MHz in Windows, as is the default behavior. So, unfortunately I doubt this particular setting affects you while being up and running in Windows.

Originally Posted by pegnose

Cool, thanks for your info!!

No problem, keep us posted!

[pegnose]

Originally Posted by Brunnis

I'm not sure exactly when the CPU standby voltage is applied, but it's not the voltage used when idling in Windows. Even though I've set the standby voltage to 1.0V, the CPU is idling at 0.8V and 800MHz in Windows, as is the default behavior. So, unfortunately I doubt this particular setting affects you while being up and running in Windows.

-.-

[s1riker]

Well just to add to the fun ...

After running HCI memtest for 48 hours with a clean bill of health and good temps all around, I figured, ok RAM's gotta be stable, so I stopped it.

I then proceeded to work all day on the PC yesterday without issue, left for a while, came back to play some GTA (maybe 20 mins at most), and left it again. Went to check on it this morning and it was frozen again.

So C-State 8 disabled, increased VCCIO and load line calibration, and new bios with MC 6A, made no difference.

EDIT: and for chalsall, I know this is not the scientific method, I'm trying multiple things at a time. But the problem is I have only 1 workstation and I need to use it as well for real work, if I were to try 1 single thing at a time, I could be at this for years. I did however test new MB (same model), new RAM (same model) and new PSU (different model) in isolation and am confident none of those are defective.

[pegnose]

Originally Posted by s1riker

Well just to add to the fun ...

After running HCI memtest for 48 hours with a clean bill of health and good temps all around, I figured, ok RAM's gotta be stable, so I stopped it.

I then proceeded to work all day on the PC yesterday without issue, left for a while, came back to play some GTA (maybe 20 mins at most), and left it again. Went to check on it this morning and it was frozen again.

So C-State 8 disabled, increased VCCIO and load line calibration, and new bios with MC 6A, made no difference.

EDIT: and for chalsall, I know this is not the scientific method, I'm trying multiple things at a time. But the problem is I have only 1 workstation and I need to use it as well for real work, if I were to try 1 single thing at a time, I could be at this for years. I did however test new MB (same model), new RAM (same model) and new PSU (different model) in isolation and am confident none of those are defective.

Try setting all your voltages to a fixed value?